Fluke 8050 Battery Power Supply Repair Advice

[ModemHead]

I should have said I'll lift one end of C22 and CR4,  and measure current as you did.

I checked DC resistance across C21 C24 and C22 while in circuit,  which correspond to the two positive rails and-5.  There was nothing that would account for an amp of current draw. I wonder if I could have a short in a T3 winding?

I'm relatively sure you can also just lift CR4 and measure the average DC current through the break and get the about the same number.

You said the resistance on the battery terminals is 580 ohms.  You're right, this does not account for the current draw.  That suggests to me the fault is on the downstream side of a silicon junction, which isn't turned on by a low-power ohm-meter.

[tango17]

Well,  we'll learn more once I  get some measurements of current on the rails.

[ModemHead]

In a few I have to go out for the rest of the day, so the peanut gallery goes silent.

Regarding T3: This looks like a blocking type oscillator, not too much unlike your basic joule thief.  If a T3 winding were shorted, would the oscillator run at all?

[tango17]

My grateful thanks to the peanut gallery.  CR4 has 298mA read by a calibrated Fluke 87V. Freq is 5.5 kHz.
Hmmmm.
After some time flat on my back,  I'll grab some more current measurements through CR5 and CR6.

[tango17]

Recalling that I have removed the display board,  these are the currents measured:

Total battery current - 1022 mA
CR4 - 290 mA
CR5 - 5 mA
CR6 - 254 mA

Presumably the rest of the current flows from the -5 rail back to the -10 rail or via Q5/6.

Where to go from here....
1 Hunt for the missing current.
2 I haven't tried changing the selector switches to see if it affects current.
3 Try again to isolate the supply from the load to make sure it works.
4 Measure resistance between various rails,  like between +6 and -5
5 Try the heat test again and see what's getting hot from that 1+ amp
6 Remove socketed DIPs.  Many are powered +6 to -5.

I'm flat on my back for a few more hours,  so I'll keep on contemplating.

I realize I can't just add up currents,  since it's being transformed to different voltages,  but it still looks like there's some more there somewhere.  The +6 flow indicates to me that there is a heavy load on it somewhere.

[ModemHead]

I made reference DC mA measurements from my working 8050A for you. Operating normally in 20VDC mode, display board connected.  Changing ranges doesn't seem to affect the readings much.

Used AVG reading from a Fluke 87V in DC mA range, with MIN/MAX engaged.  Here the current is a 31kHz waveform, and it does vary periodically with the integration/de-integration cycling of the meter, which is why I used the AVG number.

Battery - 151.8mA @ 5.1V

CR4 - 58.3mA
CR5 -  5.2mA
CR6 - 13.8mA

Looks like you have a serious problem on the +13V and possibly both negative rails.

Yes, to get the currents to add up, you would really have to count watts, and presumably a good chunk of it is being dissipated as heat.

[tango17]

Success!

It was the LM358 on the U32 hybrid RMS converter board. I cut its power leads and the battery current dropped to 192 mA.
The rails now measure +14.6, +6.0, -5.2 and -10.2.

I don't have much luck with the U32 hybrid RMS converter board. My other 8050a had the transistor array fail. It's fixed,  but not calibrated,  and that was one of the goals when this started, to see if this unit matched the Fluke 87V and to get the repaired unit  AC calibrated.

That U32 board is supposed to be replaced as a unit, and it's hard for me to calibrate AC as I don't have a decent transfer reference.  Does a computer soundcard work for low frequencies?

[ModemHead]

Awesome! 

Anything special that led you in the direction of the LM358?

I've had really good luck with the 8050A, except for LCDs and the usual trouble with dead NiCd cells screwing things up.  In fact, the reason I have several units is because I bought "non-working" ones on eBay to get good LCDs, but they were all easy to get running.  I ended up converting two to 7-segment LEDs, in my usual gonzo sort of way.

[tango17]

Both of mine were non working,  too.  This one just needed batteries.  The other was sold as working,  but wasn't,  and after complaining, I got a full refund,  and fixed the U32.  Its display is dying,  but I've got LEDs to repair it (I've read your post on that repair).

As to what lead me there,  I figured if it's not capacitors,  it's usually an active component. I marked all the semiconductor components driven by +6 and figured it would be getting hot. I felt  them all.  None were warm,  then I remembered that the components on hybrid U32 are not shown on the schematic, and checked them. As usual, in hindsight, I should have been smarter earlier.

Now to put it back together,  replace the LM358 and regulator and do some calibration.

Thanks for the help. It's easier when you can bounce ideas off someone else.

[ModemHead]

You're welcome. I've never had to really trouble-shoot this power supply before, so this exercise caused me to stare at the schematic enough that I understand it better now.

[tango17]

Update 1: All the traces have been repaired,  the diodes resoldered,  the display board reconnected, the regulator replaced and all caps replaced (which was probably needed anyway).  The LM358 is disconnected,  but not yet replaced. I'll have to check my supplies or order a new one.  The DC is low,  but that's because I had the calibration turned way down to compensate for the high readings Iwwas seeing.  I'll do the DC calibration next. All voltages are good and it's charging and running off AC correctly.  I don't know what blew the regulator.  Probably an oops or a piece of metal on the bench shorting something.

[Jeff_Birt]

What do you do when you need to replace the batteries on one of these? I have one with the batteries missing (but have the leads) and a bad LCD. I'm assuming your rolling your own battery packs but I do not have an existing pack to replicate. I know it was taken out of service because of dead batteries and the LCD was swapped into another unit with a bad LCD. I might try an LED conversion on it if I can solve the battery issue (which I'm guessing is 3-4 NiCad batteries in parallel on each side as I seem to recall seeing something about 1.2V on the silkscreen.

[ModemHead]

Sounds like great progress.  It's fortunate you can fix the RMS module with a jellybean part.

I don't know what blew the regulator.  Probably an oops or a piece of metal on the bench shorting something.

If the LM358 was the first fault to develop, then the batteries were actually providing all the extra current.  When the batteries became disconnected, that huge load was shifted entirely to the LM317, which must have given up quickly.  Just guessing.

[ModemHead]

What do you do when you need to replace the batteries on one of these?

The original battery was two packs of two C-size NiCds, wired in series for 4.8V nominal. The power supply is designed to deliver a constant trickle charge of 140mAh with the switch off, which is a reasonable overnight C/10 charge rate for cells in the 1200 to 1500mAh capacity range of that time.  With the switch on, the charge rate is reduced to about 100mA.

I have one battery-powered unit I converted to AC-only by changing the regulator from constant-current to voltage regulation at 5.5V.  For others, I just used four generic eBay NiCd sub-C cells.  The little plastic boxes are over-sized for sub-C, so I stuffed some foam in to fill up the space.

[tango17]

I use ebay and buy 4 sub-C NiCds with tabs of at least 1200mAh each. The last set I bought was 1300 mAh, but you can go as high as will fit.  The inner dimensions of the plastic box are 53 mm by 50mm by 27mm. All sub c NiCds are 23mm in diameter,  so worry only about length and keep it under 53 mm.  You could probably fit 5/4ths sub cs into that space.

When the Option 01 8050a unit is off, but plugged in,  it puts 1.2v across a 9.1 ohm resistor as a constant current source to charge the battery pack.  That's130mA. For my batteries,  that's C/10, which is high for always on charging,  and will likely kill the batteries in a few weeks/months.  You'd likely be better off with 2400 mAh batteries,  plus they last longer if you really use it without AC. I solve that by plugging it into to a power strip that I turn on only when I'm working.  It's kept them alive.

Update 2: I had added all the U5 jumpers in the original calibration attempt with low voltages.  Now that it's repaired,  I followed the steps to measure voltage and cut the U5 jumpers to get the R11 and R12 trimpots into calibration range.  With the correct voltages,  lo and behold,  the jumpers it told me to cut were the same as those I'd added. It was still in the same range (out of 16). Further,  after calibrating with R11 at the 1.9000V range,  the R12 190.00 mV calibration (which I hadn't touched)  was at 190.01mV. Not bad.

[tango17]

.  With the switch on, the charge rate is reduced to about 100mA.

The power on switch adds 12 ohms in parallel with the 9.1 ohm,  so output current from the regulator goes up to about 230mA by my calculation.  My unit was drawing 190mA without the display board,  so charge current into the batteries might be even less than 40 mA on mine.  If the batteries are charged,  and you're going to leave it plugged in,  the batteries might last longer if you keep tho 8050 power switch on rather than off and constantly pumping 130-140 mA into the batteries  for months on end.

[Jeff_Birt]

Thanks for the tips on which batteries to obtain and the link to bypassing them. I'm still undecided what to do with this meter but converting it to LED sounds fun

[ModemHead]

If the batteries are charged,  and you're going to leave it plugged in,  the batteries might last longer if you keep tho 8050 power switch on rather than off and constantly pumping 130-140 mA into the batteries  for months on end.

The unit I was measuring upthread seems to be on the low end of current consumption.  Possibly different board revisions.  With the switch on, whatever current is left over will charge the batteries obviously.

The unit I use all the time is an LED-converted one that averages 200mA consumption with a 40mA charge rate with the switch on.  I leave it powered on 24/7 unless I'm away from home, in which case it is unplugged.  The batteries have lasted for years this way.  It is sort of handy in a power outage, it will be sitting there glowing red, ready to check batteries in flashlights.

[tango17]

Thanks for the tips on which batteries to obtain...... converting it to LED sounds fun

You can get 2200 mAh on eBay for under $2 each,  and 2400mAh for a bit more. Mine only cost $1.50, but are smaller (I needed them for another project.)  I was pretty pleased with my 8050 purchases,  two units for $36 plus $6 for a ser of batteries. One LCD display is dying,  so I'm also going to do  the LED conversion soon.

Doed anyone have any suggestions for a cheap way to do the AC calibration?  I need a stable 200Hz AC source at 2 volts.  It also calls for 10kHz 19 volts and 100 volts,  but I can live without calibration of those ranges. 

[ModemHead]

Doed anyone have any suggestions for a cheap way to do the AC calibration?  I need a stable 200Hz AC source at 2 volts.  It also calls for 10kHz 19 volts and 100 volts,  but I can live without calibration of those ranges.

For tweaking the low-end AC, I found my cheap GwInstek DDS function generator becomes fairly amplitude-stable if I let it warm up a couple hours.  Also I did some experiments and found that a computer sound card playing a sine-wave tone was quite stable.  I have one of those function-generator apps on my Android tablet, but I haven't really checked it out yet.

Mine were reasonably well-calibrated when I got them running, so I just left the adjustments alone for the high ranges and the frequency compensation.  Indeed I found that simultaneous monitoring of mains voltage agrees with my better meters, so I'm happy enough with that.

I seem to recall there have been threads on this forum before about generating stable AC for calibration.  There are some volt-nuts hanging around here.  I am only a volt-kumquat.

[tango17]

I've got a sound card,  so I could use that,  although I'd need to amplify it to get 1.9 volts it recommends for calibration.  I also remember I've got an old crystal controlled oscilloscope calibrator that runs from 10 Hz to 100 kHz in steps of 1-2-5 and with outputs from 10mV to 10V. It's square wave,  but I suppose I could try it,  or build a filter to get closer to sine wave.

[tango17]

The sound card idea works great.  I found an android app that simulates a 2 channel signal generator called Function Generator that does what I need. 

[ModemHead]

Thanks for the feedback and updates, it's good to document what the outcomes are in these threads, for future readers.

[tango17]

Well,  I haven't really updated what happened,  because I didn't have an LM358A.  I only had LM358s,  so I need to place an order, and I hate to pay $6 shipping for a $0.42 part.  I'm putting together a larger order for some other projects and will post on the final results..... eventually. However, here's an interesting tidbit from the manual errata page I'd never seen.

[ModemHead]

Never noticed that either.  Bizarre little bug. I'm away from home again, but when I get back I'm going to check that out.